The Specialized Nature of Intelligence, Civilization as AI, and Science as a Superhuman, Self-Improving System

This podcast clip delves into a nuanced definition of intelligence, positing that all intelligence, including human intelligence, is inherently specialized rather than purely general. It argues that human intelligence is uniquely adapted to the "human experience," characterized by innate priors, goal-driven behavior, and a natural aptitude for learning specific things like language. However, this specialization also implies limitations, particularly in handling very long-term problems or possessing extensive working memory, as human experience is fundamentally short-lived.

The discussion then expands the concept of intelligence beyond the individual brain, introducing "civilization" as a form of distributed, problem-solving artificial intelligence. This collective intelligence operates not on a single brain, but on a vast network of human brains, augmented by infrastructure such as books, computers, the internet, and human institutions. This broader system is capable of tackling problems on a much grander scale than any individual human, challenging the traditional hierarchical view of intelligence centered solely on a singular agent.

A significant point made is the concept of "externalized intelligence," where human cognition is offloaded and distributed into external tools and systems. Examples include writing notes, creating computer programs, books, the internet, and language itself. This externalization blurs the boundaries of what constitutes an intelligent agent, emphasizing that intelligence is deeply contextual and not confined to a single biological or computational entity. The idea of an "intelligence explosion" is also addressed, with the speaker suggesting that while exponential growth in specific task categories is plausible, the focus should shift from isolated agents to real-world, recursively self-improving systems.

Finally, the conversation highlights "science" as the closest existing example of a recursively self-improving, superhuman AI. Science, as a problem-solving and knowledge-generation system, continuously experiences the world, understands it, and acts upon it. Its self-improving nature is evident as scientific progress feeds into technological advancements, which in turn provide better tools and instrumentation, accelerating further scientific discovery. This offers a grounded perspective on how superhuman intelligence might manifest and evolve, not as a singular entity, but as a collective, evolving system.

"all intelligence is specialized intelligence even human intelligence has some degree of generality when all intelligence systems have some degree of generality they're always specialized in in one category of problems"

"the human intelligence is specialized in the human experience"

"we are basically hard-coded to learn them and we are specialized in solving certain kinds of problem and we are quite useless when it comes to other kinds of problems"

"we have this thing called civilization right which is itself a sort of problem-solving system a sort of artificial intelligence system right and it's not running on one brain is ring on network of brains"

"science as a system as an institution is a kanafeh artificially intelligent problem-solving algorithm that is superhuman"

"a lot of our intelligence is externalized when you write down some notes there is externalized intelligence when you write the computer program you are externalizing cognition"

"there is no like hardly limitation of what makes an intelligent agent it's all about context"

"science is probably the closest thing we have today to a reclusive yourself improving super human AI"

"science feeds into technology technology can be used to build better tools better computers better instrumentation and so on which in turn can make sense faster"